chatcast

• 

  Thumbnail

• 

  Main page

• 

  Mobile version

• 

  Kanban

Inspiration

The unique feature that differentiate Twitch from other forms of socials is ✨chat interaction ✨.

But often many messages go unnoticed, and chat itself can become a distraction for a streamer.

What it does

🤖 chatcast gives chat insights during live stream using AI. Such as questions, jokes and ongoing discussions.

That way, streamers can keep up with viewers and interact more.

How I built it

Frontend - Vite, Backend - Go, AI - OpenAI, DB - Postgres

Challenges I ran into

I didn't know Go when I started doing this project. So it was a fun challenge to learn it in short time.

Accomplishments that I'm proud of

Front page is a demo that works 24/7 with different streamers, so everybody can get a sense how it works right away

What I learned

Go, Goroutines, Twitch Authentication, EventPub, WebSocket processing

What's next for chatcast

Short term - Test if this product will give value to streamers and think of monetization.

Long term - Train own LLM based on open source Llama. To increase quality of results and reduce maintenance costs. Try different methods of Sentiment, Emotion and Takeaways Analysis.

I would love to hear feedback from you! ❤️

You can reach me at kirill@kachalov.xyz or eX Twitter

Built With

• go
• openai
• postgresql
• vite